0801.0288/analysis.tex
1: \section{THE OBSERVATIONS}
2: %=========================
3: \label{sec:data}
4: %The targets were observed w/ XMM on date, date and date.
5: %time-scale.  AOB.
6: %The relic galaxies we observed are 0917+75, 1401$-$33 and 1253+275.
7: %Each of the targets---0917+75, 1401$-$33 and 1253+275---were observed
8: % with \emph{XMM-Newton} in a $\sim 20$\,ksec pointing.
9: We observed each of the targets, 0917+75 and 1401$-$33,
10:  in a $\sim 20$\,ks pointing with \emph{XMM-Newton}.
11: %Radio images of the targets at 1.4\,GHz were obtained from the
12: We obtained radio images of the targets at 1.4\,GHz from the
13:  NRAO VLA Sky Survey (NVSS) catalogue \citep{nvss},
14:  via the \emph{SkyView} facility
15: % \citep*{skyview}\footnote{\url{http://skyview.gsfc.nasa.gov/}}.
16:  \citep*{skyview}\footnote{http://skyview.gsfc.nasa.gov/}.
17: %We also adopted radio parameters from \citet{HarrisSternWillis1993}
18: We also used published data from \citet{HarrisSternWillis1993}
19: %, \citet{HarrisWillisDewdney1995},
20:  and \citet{GossMcadamWellington1987}.
21: %Parameters of the observation and data are listed in Table~\ref{tab:parameters}.
22: Table~\ref{tab:source param} lists some properties of the targets
23:  and the parameters of the observations.
24: %\input{tab10parameters}
25: %\input{tab00everything}
26: %\input{tab02obsParam}
27: %\placetable{tab:source param}
28: \input{tab1}
29: 
30: %\begin{table}
31: %\label{tab:parameters}
32: %\begin{tabular}{lrrr}
33: %Source			& 0917+75	& 1401$-$33	& 1253+275 \\
34: %Redshift, $z$		& 0.125		& 0.0136	& 0.023 \\
35: %Angular size		& $4'\times8'$	& $9'\times20'$ & $10'\times$30'$ \\
36: %\end{tabular}
37: %\end{table}
38: 
39: 
40: \section{DATA ANALYSIS}
41: %======================
42: \label{sec:analysis}
43: %A quick visual inspection of the images in the \emph{XMM} Pipeline Products
44: % revealed no detection.
45: %Thus, additional measures, such as restricting the energy band
46: % to maximize the signal-to-noise ratio, were necessary.
47: %Each of the three targets---0917+75, 1401$-$33, and 1253+275---were observed
48: % with XMM in a 20\,ksec pointing.
49: We first reprocessed the X-ray data sets using the routines
50:  \emph{emproc} and \emph{epproc}
51:  in the \emph{XMM-Newton} Science Analysis Software (SAS), version 6.5.0.
52: Then, we filtered the processed data based on temporal, spectral,
53:  and spatial criteria, as described below.
54: 
55: 
56: \subsection{Event filtering}
57: %---------------------------
58: \label{sec:filtering}
59: For temporal filtering, we generated good-time intervals (GTIs)
60:  based on the double-filtering technique
61:  described in Appendix~A of \citet*{Nevalainen}:
62: we inspected
63:  light curves of high-energy events (above 9.5\,keV for MOS and 10.0\,keV for PN)
64:  of 1-ks resolution,
65:  and screened out time intervals where the high-energy count rate at any
66:  of the three detectors exceeded 120 per cent of the mean level.
67: We also inspected light curves of events within 1--5\,keV
68:  at the periphery of the field of view
69:  ($12~\mathrm{arcmin} < r < 15~\mathrm{arcmin}$,
70:   $r$ being the distance from the nominal pointing direction),
71:  and screened out time intervals where any of the count rates,
72:  summed over the periphery annulus, exceeded
73:  mean+0.028~counts~s$^{-1}$ for MOS and mean+0.056~counts~s$^{-1}$ for PN.
74: The final GTI for each observation is the intersection of the GTIs
75:  from the high- and low-energy filters.
76: It is about 89 per cent the length of the entire observation for 0917+75,
77: and about 80 per cent for 1401$-$33.
78: Table~\ref{tab:resultsCalculated} lists the temporal filtering parameters
79:  in detail.
80: %\placetable{tab:resultsCalculated}
81: %We found that only the data set for 1253+275 required flare filtering.
82: %(See Figs.?)
83: %We found that the background level was constant in both data sets,
84: % and flares were absent.
85: %\footnote{Quantify.}
86: %Nevertheless, we defined good time intervals based on the criteria described
87: % by \citet{Nevalainen}.
88: %\footnote{Did I actually do that for these two sets?}
89: \input{tab2}
90: 
91: % to restrict the X-ray flux measurements to within the extent of each target,
92: % as observed at radio frequencies,
93: % and to screen out contributions by X-ray point sources that overlap
94: % with the extended targets.
95: We restricted the region for photon extraction
96:  to coincide with the radio source extent,
97:  and we filtered out regions containing point source detections.
98: %The purpose of our spatial filtering step was
99: %Radio image of each target at 1.4\,GHz was obtained from the NVSS catalogue,
100: % via SkyView (\url{http://skyview.gsfc.nasa.gov}).
101: From the NVSS radio image of each target, we made a mask
102:  by selecting only pixels in the vicinity of the target
103:  whose flux densities are at least 5$\sigma$ above the background level
104:  in each image.
105: %\footnote{Actually, is this a problem?  If we underestimate the size
106: % of the radio emitting region, then we would underestimate the X-ray flux,
107: % thus overestimate the magnetic field, thus making our results invalid.
108: % Yet, of course, the spatial extent at this high a frequency is irrelevant
109: % anyway, at least for 0917+75.}
110: To mask out X-ray point sources in each field, we took the three source lists
111:  (one from each detector on \emph{XMM-Newton})
112:  provided with the Pipeline Products,
113:  and ran them through the SAS routine \emph{region}.
114: This produced three FITS region files for each field,
115:  specifying regions that excluded all the detected point sources,
116:  with the size of each excluded region determined by the level of flux
117:  for that particular source.
118: We combined the radio mask and X-ray point source regions
119:  to form the spatial filtering expression for each of the targets.
120: %(See Figs.?)
121: These masks are shown in Figs.~\ref{fig:0917+75} and~\ref{fig:1401-33}.
122: %\placefigure{fig:0917+75}
123: %\placefigure{fig:1401-33}
124: %\begin{figure}
125: %\label{fig:0917+75}
126: %\plotone{image0917+75.eps}
127: %\caption{XMM-Newton and NVSS images of 0917+75}
128: %\end{figure}
129: %\begin{figure}
130: %\label{fig:1401-33}
131: %\plotone{image1401-33.eps}
132: %\caption{XMM-Newton and NVSS images of 1401$-$33}
133: %\end{figure}
134: \begin{figure*}
135: %\plotone{image0917+75.eps}
136: %\plotone{f1.eps}
137: %\includegraphics[width=0.5\textwidth]{f1.eps}
138: \includegraphics[width=\textwidth]{f1}
139: \caption{\emph{XMM-Newton} (a) and NVSS (b) images of 0917+75.
140:  The two images have the same spatial extent.
141:  The source (centre) and background (annulus) regions are
142:   indicated by solid lines in (a);
143:   the background region is also shown in (b) by dotted lines.
144:  X-ray point sources detected within the \emph{XMM-Newton} field of view are
145:   marked by solid lines in (b), with the size of each region proportional
146:   to the X-ray flux of the point source.
147:  The boundary of the source region is
148:   identical to the mean+5$\sigma$ contour in the NVSS image,
149:   excluding all overlapping X-ray point source regions.
150:  Image (a) is made by
151:   (1) scaling the filtered MOS (0.2--3.7\,keV) and PN (0.2--2.3\,keV) images,
152:   according to each instrument's response to the same source model and flux,
153:   (2) correcting the scaled images for exposure and combining them
154:   with the SAS routine \emph{emosaic}, and
155:   (3) smoothing the combined image with a Gaussian function
156:   of width $\sigma = 4$~arcsec,
157:   about the size of the instrument's point spread function.
158: \label{fig:0917+75}}
159: \end{figure*}
160: \begin{figure*}
161: %\plotone{image1401-33.eps}
162: %\plotone{f2.eps}
163: %\includegraphics[width=0.5\textwidth]{f2.eps}
164: \includegraphics[width=\textwidth]{f2}
165: \caption{\emph{XMM-Newton} (a) and NVSS (b) images of 1401$-$33.
166:  See Fig.~\ref{fig:0917+75} for a description of the various features.
167:  In addition to the source and background regions
168:   at the centre and top of the images, respectively,
169:   we also mark the extent of the radio emission at 330\,MHz,
170:   according to fig.~1 in \citet{Subrahmanyan2003}.
171:  Note that the source and background regions here are equidistant
172: %  (in the plane of the sky) from the bright galaxy NGC~5419
173: %  at the centre of the Abell~S753 cluster, seen here in both images
174:   from the bright galaxy NGC~5419 at the centre of the Abell~S753 cluster,
175:   about 10~arcmin to the west (right) of the source region.
176:  Image (a) combines the filtered MOS images of 0.2--6.5\,keV
177:   and the filtered PN image of 0.2--4.1\,keV.
178: % Note also that at lower radio frequencies (100s of MHz),
179: %  the radio emission of 1401$-$33 extends substantially to the southwest.
180: % See \citet{Subrahmanyan2003} for details.
181: \label{fig:1401-33}}
182: \end{figure*}
183: 
184: % spectral filtering...
185: %The objective of spectral filtering was
186: % to maximize the signal-to-noise ratio, and thus the likelihood of a detection.
187: %In order to maximize the likelihood of a detection,
188: We determined the optimal energy band for signal extraction
189: % that maximizes the signal-to-noise ratio for each source,
190:  by comparing (1) the expected source flux density at equipartition
191: % and the average background flux density for \emph{XMM-Newton}
192:  and (2) the average background for \emph{XMM-Newton}
193:  at various energies.
194: %The source flux level was obtained with \emph{PIMMS}, using a power-law model
195: % for each target, with the parameters listed in Table~\ref{tab:source param}.
196: To estimate the source flux density at equipartition,
197:  we followed the analytical formulation in \citet{1979MNRAS.188...25H}
198:  and deduced a power-law model for the IC/CMB emission from each source,
199:  with model parameters based on data from past radio observations
200:  \citep{HarrisSternWillis1993}.
201: Tables~\ref{tab:source param} and~\ref{tab:resultsInferred} lists
202:  these parameters.
203: %\placetable{tab:resultsInferred}
204: %The background level was obtained from blank sky data sets provided
205: % on the `Background Analysis' pages at the XMM Science Centre web site.\cite{XMMbackground}
206: %For the background flux density, we used blank sky event files
207: To estimate the background, we took blank sky event files
208:  obtained from the web site of the \emph{XMM-Newton} Science Operation
209: % Centre\footnote{\url{http://xmm.vilspa.esa.es/external/xmm\_sw\_cal/background/blank\_sky.shtml}} as representative background measurements,
210:  Centre\footnote{\scriptsize http://xmm.vilspa.esa.es/external/xmm\_sw\_cal/background/blank\_sky.shtml} as representative background measurements,
211:  and computed a background spectrum from the files.
212: \citet{CarterRead2007} described these blank sky data sets in detail.
213: %The expected signal-to-noise ratios were computed for energy bands with
214: % maximum and minimum energy values at increments of 0.1\,keV.
215: %The band with the highest signal-to-noise ratio for each source spectrum
216: % and detector background was selected.
217: With these source and background spectra, we selected an energy band
218:  $[E_\mathit{min}, E_\mathit{max}]$ and calculated the signal-to-noise
219:  ratio (S/N) associated with the band.
220: We then varied $E_\mathit{min}$ and $E_\mathit{max}$ at 0.1-keV increments
221:  until we obtained the energy band with the maximum S/N.
222: Using this method, we arrived at the optimal MOS bands
223:  0.2--3.7\,keV for 0917+75 and 0.2--6.5\,keV for 1401$-$33.
224: Table~\ref{tab:resultsCalculated} shows the selected energy bands
225:  for PN as well.
226: 
227: %In addition to the GTIs and energy bands obtained from above,
228: In addition to the GTIs and energy bands obtained as described above,
229:  we further applied the following event filtering expressions:
230: % (\verb$#XMMEA_EM && PATTERN <= 4$ for MOS
231: % and \verb$#XMMEA_EP && PATTERN <= 12$ for PN).
232: % (\verb$FLAG == 0 && PATTERN <= 12$) for MOS
233: % and (\verb$FLAG == 0 && PATTERN <= 4$) for PN.
234:  (\texttt{FLAG == 0 \&\& PATTERN <= 12}) for MOS
235:  and (\texttt{FLAG == 0 \&\& PATTERN <= 4}) for PN.
236: These expressions retained only events triggering 1 or 2 pixels,
237:  and only those flagged as valid by the standard processing routines.
238: From the filtered event list, we generated images and exposure maps.
239: We then used the masks and regions described above to select
240:  the region of sky corresponding to each target.
241: %The photon count rate is obtained as the quotient of the number of counts
242: % in the selected region divided by the sum of the corresponding
243: % exposure map values.
244: The total counts divided by the mean of the exposure map in the selected region
245:  gave the event rate.
246: % gave the photon count rate;
247: 
248: 
249: %\section{RESULTS}
250: %-----------------
251: %\label{sec:results}
252: \iffalse
253: \footnote{Per Fiona's advice, I substantially reorganised this section.
254: The outline in typewriter text above is just FYI, and will be removed
255:  before submitting.}
256: \scriptsize
257: \begin{verbatim}
258: %1. Spell out numerical results.  Conclusion: No detection.
259: %2. Constaints: How to get background sigma; spell out sigma; compare w/ src;
260: %		B limit; conclusion: B is higher than equipartition.
261: %3. 0917+75 background: a. ACE data; b. blank sky data; c. light curves.
262: %		Conclusion: clueless, but non-detection OK, affects limits.
263: %4. 1401-33 background: a. thermal emission; b. vignetting
264: %		Conclusion: imperfect, but we've done all we can.
265: \end{verbatim}
266: \normalsize
267: \fi
268: \subsection{Background determination and flux upper limit analysis}
269: %------------------------------------------------------------------
270: \label{sec:background}
271: %The background measurements for both observations require additional discussion.
272: The two observations in this paper pose very different challenges
273:  for background determination.  We discuss our approach for each observation
274:  seperately in the following.
275: 
276: \subsubsection{0917+75}
277: %^^^^^^^^^^^^^^^^^^^^^^
278: For 0917+75, the time of our observation unfortunately coincided
279:  with a coronal mass ejection (CME) from the Sun.
280: \citet*{Snowden} reported an enhancement in the \emph{XMM-Newton} background,
281:  especially at 0.5--1.0\,keV,
282:  concurrent with an enhancement in the solar wind
283:  measured by the \emph{Advanced Composition Explorer} (\emph{ACE})
284:  and other monitoring spacecraft.
285: When we obtained data
286: % from \emph{ACE}\footnote{\url{http://www.srl.caltech.edu/ace/ASC/}}
287:  from \emph{ACE}\footnote{http://www.srl.caltech.edu/ace/ASC/}
288:  and inspected the light curves 
289:  at the time of our observations (Fig.~\ref{fig:ace}),
290:  we found that our observation of 0917+75 coincided with an episode
291:  of significant enhancement of the O$^{7+}$/O$^{6+}$ ratio in the solar wind,
292:  characteristic of a CME (R.~C.~Ogliore 2006, private communication).
293: \begin{figure}
294: %\plotone{ace.eps}
295: %\plotone{f3.eps}
296: %\includegraphics[width=0.5\textwidth]{f3.eps}
297: \includegraphics[width=0.47\textwidth]{f3}
298: \caption{Proton flux and ratio of oxygen charge states O$^{7+}$/O$^{6+}$
299:   in the solar wind, measured by \emph{ACE} during the time of our observations.
300:  The two horizontal lines in each panel indicate the mean and 90-percentile
301:   levels of each quantity over the first 100~days of 2002.
302:  The two vertical lines in each panel delimit
303:   the period of the \emph{XMM-Newton} observation.
304:  The 64~s-averaged proton flux data are
305:   from the SWEPAM instrument on \emph{ACE},
306:   while the hourly charge state ratios are from the SWICS/SWIMS instrument.
307:  Note that the break in the O$^{7+}$/O$^{6+}$ curve
308:   at the time of our observation of 0917+75 was due to missing data.
309:  Yet, from the values before and after the break,
310:   one can infer that the O$^{7+}$/O$^{6+}$ ratio during the observation
311:   is probably higher than the 90-percentile.
312:  There is a $\sim$1~hour travel time by the solar wind from the L1 point,
313:   the location of \emph{ACE}, to the Earth.
314:  The time shown here is the time measured by \emph{ACE}.
315: \label{fig:ace}}
316: \end{figure}
317: 
318: To assess the impact of this event on the background level in our observation,
319: % we repeated the background measurement using two sets
320: % of blank sky event files---from \citet{Nevalainen} and \citet{CarterRead2007},
321: % respectively---which serve as independent, standard data sets.
322:  we compared the event rate in our data set with the rates
323:  in two standard blank-sky data sets independent of our observation:
324:  one set was from \citet{Nevalainen} and the other from \citet{CarterRead2007}.
325: % that we used for spectral filtering (see \S\ref{sec:filtering}),
326: We chose these two sets because we employed the flare-filtering recipe
327:  of \citet{Nevalainen} in this study, while \citet{CarterRead2007} provides
328:  a much larger data set, with better event statistics.
329: We refiltered the \citeauthor{CarterRead2007} data set
330:  with the (\texttt{FLAG == 0}) expression
331:  so that comparison of the three data sets is consistent.
332: However, we have not refiltered the \citeauthor{CarterRead2007} data set
333:  for flares using the recipe in \citet{Nevalainen},
334:  as its full-view hard-band event rates
335:  (0.26 and 2.7~counts~s$^{-1}$ for MOS and PN, respectively)
336:  are above the range deemed consistent with the sample of \citet{Nevalainen},
337:  making the hard-band filtering method inapplicable.
338: To minimize the variability of any other systematic parameters
339:  (such as vignetting),
340:  we applied the masks and regions that we obtained
341:  for the source regions of 0917+75 in this blank-sky measurement,
342:  thus selecting the same detector pixels.
343: We computed count rates at various energies
344:  from the filtered blank sky event files,
345:  and compared them with our source-region count rates for 0917+75,
346:  both without exposure map correction.
347: Fig.~\ref{fig:0917+75spectrum} shows spectra of 0917+75
348:  and of the blank sky data thus obtained.
349: \begin{figure}
350: %\plotone{spectrum0917+75.eps}
351: %\plotone{f4.eps}
352: %\includegraphics[width=0.5\textwidth]{f4.eps}
353: \includegraphics[width=0.47\textwidth]{f4}
354: \caption{Spectra of 0917+75 and of the blank sky data.
355:  The solid line shows the spectrum of 0917+75 from MOS2,
356:   convolved with a 100~eV-wide boxcar.
357:  The dashed and dash-dot lines show the blank sky spectra
358:   (also convolved with a 100~eV-wide boxcar)
359:   from \citet{Nevalainen} and from \citet{CarterRead2007}, respectively.
360:  Both blank sky spectra contain events
361:   extracted from the same detector coordinates of MOS2
362:   as the source region in this study,
363:   and scaled to the spectrum of 0917+75
364:   by equating the GTI-filtered livetimes % length of the GTIs and livetimes.
365:  Data from \citet{CarterRead2007} is also refiltered with (\texttt{FLAG == 0})
366:   for consistent comparison.
367: %  and the count rates above 9.5~keV, respectively,
368:   in the data sets.
369:  Note the excess of counts at discrete `spectral lines' at 0.5--1.0~keV
370:   when compared to the data from \citet{Nevalainen},
371:   and the overall deficit of counts when compared to \citet{CarterRead2007}
372: % Note the deficit of counts in the source spectrum,
373: %  compared to the blank sky spectrum in either normalization scheme.
374: % Note the elevated count rate in the source spectrum at 0.5--1.0~keV
375:   (see text for an explanation).
376: % The count rates displayed here are not corrected for dead time.
377:  The count rates displayed here are livetime corrected.
378: \label{fig:0917+75spectrum}}
379: \end{figure}
380: %It shows that the continuum level is lower for 0917+75
381: % than for the blank sky data.
382: %The result is against our expectation---despite evidence of a CME
383: % from the \emph{ACE} data, the source event rates for 0917+75 are
384: % still lower than the blank sky rates from the same regions of the detectors.
385: It shows that the continuum level in our observation of 0917+75 is consistent
386:  with that from \citet{Nevalainen}, but lower than the level from
387:  \citet{CarterRead2007}.
388: We believe the difference between the two blank sky data sets is
389:  due to different levels of flare filtering.
390: As we employed the more stringent double-GTI filtering recipe
391:  of \citet{Nevalainen} in this study, our spectrum of 0917+75 should be
392:  directly comparable with a spectrum of the same detector region made
393:  from their data set.
394: We note that the CME during our observation explains the excess line emission
395:  (relative to the continuum) at multiple energies below~1.0\,keV
396:  in the spectrum of 0917+75, which is not seen in either blank sky spectrum.
397: 
398: To further investigate this background discrepancy, we looked at light curves
399:  generated from our observation of 0917+75.
400: Fig.~\ref{fig:0917+75light curve} shows the light curves of all events
401:  above about 10\,keV from the entire field of view, which we generated
402:  to determine the high-energy GTIs (see \S\ref{sec:filtering}).
403: \begin{figure*}
404: %\plotone{lightcurves0917+75.eps}
405: %\plotone{f5.eps}
406: \includegraphics[width=\textwidth]{f5.eps}
407: \caption{Light curves of high-energy events (left)
408:   and of events at the periphery of the field of view (right)
409:   during the observation of 0917+75.
410:  The three rows of graphs show, from top to bottom,
411:   events from MOS1, MOS2 and PN, respectively.
412:  The abscissae show the time since the start of the observation, in seconds.
413:  The ordinates show the count rate, in counts~s$^{-1}$.
414:  The three solid lines in each graph on the left indicate
415:   the mean and (mean$\pm$20\%) high-energy count rates.
416:  The long-dash lines show the same quantities with the median.
417:  The two short-dash lines delimit the `normal' range of count rates
418:   reported by \citet{Nevalainen},
419:   which is exceeded during most of this observation.
420:  On the right, the solid (long-dash) lines delimit the acceptable range
421:   of count rates computed with the mean (median),
422:   according to \citet{Nevalainen}.
423:  The short-dash lines delimit the `normal' range
424:   observed by \citet{Nevalainen};
425:   only the top line is seen in the top two graphs.
426: \label{fig:0917+75light curve}}
427: \end{figure*}
428: %Although the count rates within the optimal energy band of 0.2--3.7\,keV were
429: % lower for 0917+75 than for and the blank sky data,
430: This figure shows that
431: % the count rates over the entire field of view above about 10\,keV,
432: % as shown in these light curves, were above the `acceptable' range
433:  the count rates over the entire field of view above about 10\,keV
434:  were above the `acceptable' range
435:  of high-energy particle background rate reported by \citet{Nevalainen}
436:  for the majority of our observation period, and for all three detectors.
437: This is evidence that our entire observation was
438:  plagued with low-amplitude long-duration flares,
439:  and that residual flares still lurk in the background
440:  after our double-GTI filtering.
441: Thus, these light curves are consistent with the \emph{ACE} data,
442:  suggesting an elevated background.
443: 
444: %To determine background count rates, we first attempted to use
445: % the blank sky event lists provided by the XMM-SSC, but ran into
446: % several problems.  Firstly, the XMM-SSC only provides corresponding
447: % exposure maps for 2--10\,keV.  This energy band is not desirable
448: % for us, because the targets of our observations are all power-law
449: % sources with flux that falls off substantially at high energies.
450: %A quick visual inspection of the Pipeline Products revealed no detection;
451: % this implies that maximizing the signal-to-noise ratio by restricting
452: % the energy band is essential.
453: %Also, by comparing spectra generated from the blank sky event lists
454: % and from our data, using the same spatial filtering criteria in
455: % detector coordinates in both cases, we found a mismatch in the two data sets:
456: %The silicon and aluminium instrument lines from the two sets do not appear
457: % at the same pulse height invariant values.  Also, the general shape of
458: % the continuum levels, especially at very low energies, are different.
459: %(See Figs.?)
460: %We suspect that these differences were at least partly attributable to
461: % different versions of SAS software used to generate the two data sets.
462: %(According to comments in the header units of the blank sky event lists,
463: % these files were generated with XMM-SAS version~6.6.0, dated 2005/12/07 18:03,
464: % which seems to be an internal release version that is not made available
465: % to the public).  In any case, due to these inconsistencies, we decided
466: % against determining the background count rate from the blank sky event lists,
467: % and we turned out attention to obtaining local background regions for each
468: % observation instead.
469: 
470: %To determine background count rates, we opted to consider local background
471: % regions, as blank sky event lists with exposure maps for arbitrary energy
472: % bands were unavailable.
473: %Also, each of our observations has its own problems that make
474: % the use of blank sky event lists unsuitable (see \S\ref{sec:results}).
475: Because our observation was contaminated by flares,
476:  count rates from blank-sky data sets do not correctly reflect
477:  the nature of the background in our data.
478: Besides, according to Fig.~\ref{fig:0917+75spectrum},
479:  the source rate in our data does not exceed the blank-sky count rates.
480: % in other words, we have a non-detection.
481: %Thus, in addition to the background level itself,
482: % we need to measure its \emph{variation} in order to set a flux limit
483: % we need to measure its variation in order to set a flux limit
484: % on 0917+75.
485: %For these two reasons, we considered multiple local background regions.
486: Both of these reasons make it impossible for us to use blank-sky data
487:  for quantitative background subtraction.
488: Thus, we measured the background level and its variation
489:  (for setting a flux limit in the case of non-detection)
490:  solely from multiple local background regions.
491: We obtained the local background regions using a similar approach as we did
492:  for the source regions,
493:  applying temporal and spectral filters
494:  described in \S\ref{sec:filtering}.
495: For the spatial filters, we inspected the radio image of 0917+75 from NVSS,
496: % of each \emph{XMM-Newton} pointing,
497:  as well as X-ray images from the Pipeline Products,
498:  to find regions of sky within the \emph{XMM-Newton} field of view
499: % where the radio flux is low,
500:  but securely outside the target, as seen in radio frequencies,
501:  and where there are few detectable X-ray point sources.
502: %We first drew these regions by hand, using the polygon region tool in ds9,
503: % over the radio and X-ray images.
504: %For 0917+75, we selected two background regions:
505: %The first one was an annulus immediately outside the target,
506: %For 0917+75,
507: We selected an annulus immediately outside the target,
508:  with area similar to the source region.
509: This annulus is inscribed within the central CCDs of MOS1 and MOS2,
510: % which simplifies background subtraction,
511:  which simplifies the comparison between the source and background regions,
512:  as potentially different background levels at different CCDs
513:  \citep{PradasKerp2005} become irrelevant for the two MOS modules.
514: %The second background region was a polygon further to the north of the target,
515: % with low radio flux and few X-ray point sources.
516: With the background region thus selected,
517:  we excluded X-ray point sources from the region in the same way
518:  as was done for the source region.
519: Fig.~\ref{fig:0917+75} shows the resulting background annulus.
520: We then applied this local background region to the filtered images and
521:  exposure maps described in \S\ref{sec:filtering},
522:  and obtained the total background count,
523:  mean exposure, and background count rate scaled to the total exposure
524:  in the source region.
525: %To obtain an upper limit of source flux,
526: % we recalculate the background level with multiple background regions
527: % at the vicinity of the original, and obtain the standard deviation $\sigma$
528: % of this sample.
529: To assess the variation of the background level in our observation,
530:  we obtained additional background regions in the vicinity of the annulus
531:  and scaled the event rates within them in a similar fashion,
532:  and calculated the standard deviation $\sigma$ of scaled event rates
533:  in this sample.
534: 
535: To assess the amount of degradation in our data due to residual flares,
536:  we calculated the count rates of individual pointings that make up
537:  the blank sky data set of \citet{Nevalainen}.
538:  We then compared the background variation (over space) in our data
539:  with the variation of the blank-sky count rates (over time).
540: We found the variation in our background to be greater
541:  than that in the blank sky data by factors of 1.26, 1.27 and 2.05
542:  for the MOS1, MOS2 and and PN detectors, respectively.
543: %The flux limit we set in this study is worsened accordingly.
544: The flux limit we set in this study is affected accordingly.
545: 
546: \subsubsection{1401$-$33}
547: %^^^^^^^^^^^^^^^^^^^^^^^^
548: Compared to 0917+75, the nature of the background
549:  in our observation of 1401$-$33 is quite different.
550: The observation was made at a time of average solar wind activity,
551: as shown by \emph{ACE} data during this period (see Fig.~\ref{fig:ace});
552:  the overall count rates above about 10\,keV are
553:  also well within the accepted quiescent levels.
554: However, an X-ray halo of thermal emission centred at the neighbouring
555:  bright galaxy NGC~5419 about 10~arcmin away west of northwest
556:  \citep{Subrahmanyan2003}
557:  dominates the local background, producing excess flux across the entire
558:  field of view when compared to blank sky data.
559: This thermal emission is so strong that spectral lines caused by the solar wind,
560:  which are normally present below 1.0\,keV, are not observable.
561: %Given this dominating thermal component across the field of view,
562: % comparison with a standard background data set is meaningless.
563: %Instead, we made the assumption that the thermal emission is spherically
564: To assess the amount of thermal emission from the halo
565:  in our source region for 1401$-$33,
566:  we made the assumption that the thermal emission is spherically
567:  symmetric about the centre of the cluster (ie, NGC~5419),
568:  and chose a local background region equidistant from the centre
569:  of the thermal emission as the source region. % on the focal plane.
570: Our assumption is based on a previous X-ray observation of the cluster
571:  with the \emph{ROSAT} PSPC (in which IC/CMB emission from the relic was
572:  not detected); fig.~8 in \citet{Subrahmanyan2003} is an X-ray image
573:  from the observation, showing rotationally symmetric contours about NGC~5419.
574: % with about 8\% of the peak emission in both our source and background regions.
575: %\footnote{Do I need to give a number here, eg, the brightness level?}
576: %We acknowledge that such a background region is far from ideal,
577: % as our results depend heavily on the validity of the symmetry assumption.
578: %However, given the dominance of the thermal emission, this appears to be
579: % a reasonable assumption to first order.
580: The regular symmetric shape of the cluster justifies our choice
581:  of the background region.
582: %In addition, because a large portion of the X-ray field of view overlaps
583: % with the extent of the radio source at 330\,MHz \citep{Subrahmanyan2003},
584: % we chose a circular region just to the north of the target,
585: % instead of the usual annulus surrounding the source region.
586: % (but see \S\ref{sec:results} for a caveat).
587: In addition, \citet{Subrahmanyan2003} also revealed that
588:  a large portion of the \emph{XMM-Newton} field of view in our observation
589:  overlaps with the extent of the radio source at 330\,MHz.
590: Thus, we chose a circular region just to the north of the target,
591:  instead of the usual annulus surrounding the source region.
592: Fig.~\ref{fig:1401-33} shows the resulting background region,
593:  with the extent of the 330-MHz emission for reference.
594: To calculate the local background event rate and its variation,
595:  we applied the background region to the filtered image and exposure map,
596:  and obtained additional background regions, following the same recipe
597:  as outlined above for 0917+75.
598: 
599: %Another issue with our background region for 1401$-$33 is vignetting.
600: Because our background region for 1401$-$33 is not close to the centre
601:  of the telescopes' field of view, vignetting is a potential issue.
602: \citet{PradasKerp2005} reported an overcorrection of vignetting towards
603:  the rim of each detector by the routine \emph{eexpmap} that generates
604:  exposure maps in SAS versions 5.3.3 and 5.4.1.
605: %\footnote{FYI, I emailed Pradas, but got an `unknown recipient' mail error.
606: % I then emailed Kerp, but did not get a reply.}
607: To avoid overlapping with the extent of the radio emission of 1401$-$33
608:  at 330\,MHz, we have no choice but to place our background region
609:  at the periphery of the \emph{XMM-Newton} field of view.
610: Thus, the overcorrection of vignetting affects our analysis,
611:  potentially overestimating the count rate in the background region.
612: %We further discuss this issue at the end of the next section.
613: %\footnote{I could compare the counts in the peripheral polygon
614: % and annulus regions of 0917+75, assume those are equal, and infer
615: % the amount of overcorrection, but I have not done that yet, and so,
616: % I have no number to put in here.  However, the change will be relatively
617: % simple.  Probably just adding an extra sentence.  I have emailed J. Kerp
618: % in the meantime.}
619: On the other hand, the particle background that dominates at high energies
620:  is not vignetted.
621: To correct for vignetting of the local background properly,
622:  we followed a two-step background subtraction method
623:  described in Appendix~A of \citet{Arnaud2002}:
624: We calculated event rates in the blank-sky data from \citet{Nevalainen}
625:  at pixels in the source and background regions
626:  and within the optimal energies in this study.
627: We scaled these source- and background-region event rates
628:  according to the ratio of hard-band rates in our data
629:  and in the blank-sky data.
630: We then subtracted the scaled rates from the event rates in our data
631:  to remove the particle background from both the source and background regions.
632: Having eliminated the non-vignetted component,
633:  we then scaled the event rate in the background region to the source region
634:  according to the ratio of summed exposure in each region,
635:  for vignetting correction.
636: Finally, we arrived at the source event rate
637:  by substracting the scaled background-region rate from the source-region rate.
638: Table~\ref{tab:resultsCalculated} shows the relevant numbers in this analysis.
639: 
640: 
641: \subsubsection{Flux and field limits}
642: %^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
643: \label{sec:results}
644: Table~\ref{tab:resultsCalculated} summarizes our quantitative results.
645: %In the following paragraphs, we discuss only results from the MOS detectors,
646: In this paper, we discuss only results from the MOS detectors,
647:  as the MOS and PN results yield the same conclusions, and the MOS results
648:  provide more stringent limits.
649: %\placetable{tab:resultsCalculated}
650: %From these numbers, we see that the flux obtained from the two source regions
651: % is consistently lower than those from the background region
652: % in each observation, implying non-detection for all three observations.
653: % are both substantially lower than the local background levels;
654: %\footnote{Fiona, the source flux levels are 16--36$\sigma$
655: % below the scaled background levels.  So, I cannot simply say
656: % that `the source flux levels are consistent with the background levels
657: % to within N$\sigma$,' as you suggested.}
658: For 0917+75,
659: % the flux in the source region is consistent
660: % with that in the background region to within 3$\sigma$;
661: % the area-scaled event rates in the source and background regions are
662: % both 0.018~counts/sec (0.2--3.7\,keV), agreeing to better than 10\%.
663:  the event rates within the source region in the optimal band of 0.2--3.7\,keV
664:  are 0.0186 and 0.0176~counts~s$^{-1}$ for MOS1 and MOS2, respectively,
665:  while the corresponding area-scaled event rates in the background regions are
666:  $(0.0180 \pm 0.0016)$ and $(0.0194 \pm 0.0015)$~counts~s$^{-1}$, respectively.
667: %For 1401$-$33, the event rates in the on-axis source region
668: % (0.0347 and 0.0370~counts~s$^{-1}$, respectively) are
669: % only about 70 per cent of the area-scaled rates in the off-axis background region
670: % [$(0.0556 \pm 0.0041)$ and $(0.0568 \pm 0.0035)$~counts~s$^{-1}$, respectively].
671: For 1401$-$33, the blank-sky substracted event rates
672:  in the on-axis source region
673:  (0.0142 and 0.0168~counts~s$^{-1}$, respectively) are
674:  only a fraction of the blank-sky subtracted and area-scaled rates
675:  in the off-axis background region
676:  [$(0.0310 \pm 0.0043)$ and $(0.0319 \pm 0.0035)$~counts~s$^{-1}$, respectively].
677: %In other words, we did not detect the IC/CMB emission from either source.
678: These numbers, together with the appearance of the X-ray images
679:  in Figs.~\ref{fig:0917+75} and~\ref{fig:1401-33}, indicate that
680: % there is clearly not a detection of the IC/CMB emission from either source.
681:  there is not a detection of the IC/CMB emission from either source.
682: 
683: %We used PIMMS to convert the photon count rate to energy flux,
684: % and taking into account galactic absorption in this process.
685: % we converted the count rate to energy flux with PIMMS \citep{pimms},
686: % taking redshift and galactic absorption into account in the process.
687: We converted the measured event rate to the unabsorbed energy flux
688:  with PIMMS \citep{pimms}, assuming a power-law spectrum and
689:  accounting for redshift and galactic absorption.
690: We set the X-ray spectral index to be the same
691:  as the radio spectral index~\citep{1979MNRAS.188...25H},
692:  assuming any X-ray emission to be IC/CMB in origin.
693: %The assumed galactic \ion{H}{1} column densities were
694: %The assumed galactic \mbox{H\,{\sc i}} column densities were
695: We took the spectral indices ($\alpha=1.0$ for 0917+75 and 1.4 for 1401$-$33)
696:  and redshifts ($z=0.125$ and 0.0136, respectively)
697:  from previous radio measurements,
698:  and the galactic \mbox{H\,{\sc i}} column densities
699:  ($2.09 \times 10^{20}$ and $5.44 \times 10^{20}\,cm^{-2}$)
700:  from the FTOOLS programme nH.
701: % Table~\ref{tab:source param} gives their values.
702: %Note that PIMMS is designed for the study of point sources only,
703: %and it expects the input count rate to be that
704: %from a point source.\cite{PIMMSmanual}
705: %Thus, we scale the count rates measured from the extended sources
706: % by the ratio of the XMM point spread function (15'') to the size
707: % of each source, before entering the numbers in PIMMS.
708: 
709: %To obtain an upper limit of the source flux,
710: % we calculated the background variation in each observation
711: % with multiple background regions
712: % at the vicinity of the original, and obtained the standard deviation $\sigma$
713: % of this sample.
714: \iffalse
715: To assess the maximum source flux,
716:  we adopted twice the r.m.s. variation in the local background as an upper limit,
717:  to make direct comparison with previous studies possible.
718: When converted to energy flux at 0.3--10.0\,keV, the +2$\sigma$ upper limits
719:  are $3.48 \times 10^{-14}$\,erg~s$^{-1}$~cm$^{-2}$ for 0917+75
720:  and $9.71 \times 10^{-14}$\,erg~s$^{-1}$~cm$^{-2}$ for 1401$-$33.
721: The measured flux in the source region of 0917+75 is consistent with this
722:  2$\sigma$ limit, while the source flux for 1401$-$33 is as much as 5$\sigma$
723:  below the measured mean background.
724: \fi
725: To assess the maximum source flux, we adopted
726:  three times the r.m.s. variation in the local background as an upper limit.
727: When converted to energy flux at 0.3--10.0\,keV, the +3$\sigma$ upper limits
728:  are $5.22 \times 10^{-14}$\,erg~s$^{-1}$~cm$^{-2}$ for 0917+75
729:  and $1.47 \times 10^{-13}$\,erg~s$^{-1}$~cm$^{-2}$ for 1401$-$33.
730: The measured flux in the source region of 0917+75 is consistent with this
731:  3$\sigma$ limit, while the source flux for 1401$-$33 is about 4$\sigma$
732:  below the measured mean background.
733: Using Equation~11 in \citet{1979MNRAS.188...25H},
734: % Equation~9 in \citet{1998PASJ...50..389H},
735:  and radio measurements reported in \citet{HarrisSternWillis1993}
736:  and \citet{GossMcadamWellington1987},
737: % we obtained 2$\sigma$ lower limits on the magnetic field strength
738:  we obtained 3$\sigma$ lower limits on the magnetic field strength
739:  at each source;
740: % they are 0.93\,$\mu$G for 0917+75 and 1.9\,$\mu$G for 1401$-$33
741:  they are 0.81\,$\mu$G for 0917+75 and 2.2\,$\mu$G for 1401$-$33.
742: % (see Table~\ref{tab:resultsInferred} for further details).
743: %\input{tab3}
744: For comparison, the field strength obtained with the `classical' equipartition
745:  formula (with a low-frequency cutoff of the synchrotron emission
746:  at $\nu=10\,\mathrm{MHz}$)
747: % are 0.5\,$\mu$G and 0.8\,$\mu$G for 0917+75 and 1401$-$33, respectively.
748:  are 0.63\,$\mu$G and 1.3\,$\mu$G for 0917+75 and 1401$-$33, respectively.
749: Table~\ref{tab:resultsInferred} provides further details on these calculations.
750: \input{tab3}
751: Alternatively, using a revised equipartition formula
752:  from \citet*[Equation~A4]{BrunettiSettiComastri1997},
753: % with the low-energy electron population cutoff at $\gamma = 100$,
754: % the field strengths are 0.6\,$\mu$G and 1.0\,$\mu$G, respectively.
755:  with the low-energy electron population cutoff
756:  at $\gamma_\mathrm{min} = 1069 \sqrt{E_\mathrm{x,min}/\mathrm{keV}} = 500$
757:  \citep{1979MNRAS.188...25H} to match the energy range of \emph{XMM-Newton},
758:  and with $D(\delta) \sim 1$,
759:  the field strengths are 0.79\,$\mu$G and 1.6\,$\mu$G, respectively.
760: We note, however, that these values are very sensitive
761:  to the choice of $\gamma_\mathrm{min}$,
762:  whose true value is unknown without better knowledge
763:  of the low-energy synchrotron spectrum.
764: \iffalse
765: These numbers are both higher than the theoretical values computed
766:  from the equipartition formula, up to a factor of two,
767:  indicating that the magnetic fields at these sources are stronger
768:  than expected.
769: 
770: %Despite these investigations, we have not found a satisfactory explanation
771: % for the discrepancy between the blank-sky and local event rates.
772: %However, the uncertainty in the background level does not affect
773: Although it is unfortunate that residual flares increase
774:  the background level in our observation of 0917+75,
775:  this increase in the background level does not affect
776:  the primary conclusion that this observation is clearly a non-detection.
777: If anything, it makes the limits we impose on the IC/CMB emission
778:  and the magnetic field strength less stringent than those obtained
779:  from a typical \emph{XMM-Newton} observation,
780:  as elevated background level due to CME-related flares would also increase
781:  the statistical fluctuation in the background.
782: %on the other hand, if the measured background in our observation is
783: % indeed lower than that in a typical \emph{XMM-Newton} observation,
784: % it would suggest that we have neglected a (negative) systematic component,
785: % and thus overestimated the statistical fluctuation in the data.
786: %\footnote{Fiona, is this conclusion correct?  I am wary of the last statement.
787: %	However, I do not know how to explain it otherwise...
788: %	Also, how should I quantify this?  I thought I understood this
789: %	when we talked on Tuesday, but upon thinking about it further,
790: %	I do not.}
791: %However, as with the elevated background for 0917+75,
792: % we point out that this systematic uncertainty in the background level
793: % does not change the primary conclusion that our observation yields
794: % a non-detection of IC/CMB emission from 1401$-$33.
795: The same idea holds for the effect of vignetting for 1401$-$33:
796:  systematic uncertainty in the background level
797:  does not change the primary conclusion of non-detection.
798: Rather, it makes the limits we impose on the IC/CMB flux and the magnetic
799:  field less stringent.
800: \fi
801: 
802: 
803: \iffalse
804: It seems peculiar to us that for 1401$-$33, the local background flux
805: % was substantially higher than the total flux in the source region.
806:  was higher than the total flux in the source region.
807: \citet{PradasKerp2005} reported an overcorrection of vignetting towards
808:  the rim of each detector by the routine \emph{eexpmap} that generates
809:  exposure maps in SAS versions 5.3.3 and 5.4.1.
810: While we do not know if this overcorrection is still in place in version 6.5.0,
811:  which we used in this study, it would explain the high background level
812:  we measured.
813: To ascertain that our non-detection results were not due to vignetting effects,
814:  we measured count rates from the blank sky event files
815:  that we used for spectral filtering (see \S\ref{sec:filtering}),
816:  applying the same masks and regions that we obtained
817:  for the source regions above, thus selecting the same detector pixels
818:  and eliminating detector effects in principle.
819: We then compared these blank sky count rates
820:  with our source-region count rates, both without exposure map correction.
821: The result is similar for 0917+75 -- the blank sky rates are
822:  higher than the source rates from the same regions of the detectors
823:  (see Fig.~\ref{fig:0917+75spectrum}).
824: %\placefigure{fig:0917+75spectrum}
825: %The background level in this observation is further discussed
826: % in \S\ref{sec:0917+75}.
827: As for 1401$-$33,
828:  an X-ray halo of thermal emission centred at the neighbouring
829:  bright galaxy NGC~5419 about 10~arcmin away to the west of northwest
830:  \citep{Subrahmanyan2003}
831:  dominates the local background, producing excess flux across the entire
832:  field of view when compared to blank sky data.
833: Thus, we are unable to make a similar comparison as we have done for 0917+75.
834: %\begin{figure}
835: %\label{fig:0917+75spectrum}
836: %\plotone{spectrum0917+75.eps}
837: %\caption{Spectra of 0917+75 and of the blank sky data.}
838: %\end{figure}
839: 
840: The background level in the observation of 0917+75 requires
841:  additional discussion.  
842: Fig.~\ref{fig:0917+75light curve} shows the light curves of all events
843:  above about 10\,keV from the entire field of view, which we generated
844:  to determine the high-energy GTIs (see \S\ref{sec:filtering}).
845: %\placefigure{fig:0917+75light curve}
846: Although the count rates within the optimal energy band of 0.2--3.7\,keV were
847:  lower for 0917+75 than for both the local background region
848:  and the blank sky data,
849:  the count rates over the entire field of view above about 10\,keV,
850:  as shown in these light curves, were above the `acceptable' range 
851:  of high-energy particle background rate reported by \citet{Nevalainen} 
852:  for the majority of our observation period, and for all three detectors.
853: This hints at the possibility that our entire observation might have been
854:  plagued with low-amplitude long-duration flares,
855:  and that residual flares still lurk in the background
856:  after our double GTI filtering.
857: %\begin{figure}
858: %\label{fig:0917+75light curve}
859: %\plotone{lightcurves0917+75.eps}
860: %\caption{Light curves of high-energy events during the observation of 0917+75.}
861: %\end{figure}
862: 
863: A second piece of evidence comes from low energies.
864: Fig.~\ref{fig:0917+75spectrum} shows spectra of 0917+75
865:  and of the blank sky data, both without exposure correction.
866: Although the continuum level was lower for 0917+75 than for the blank sky data,
867:  there were apparently excess line emissions at multiple energies below~1.0 keV,
868:  not seen in the other spectrum.
869: \citet*{Snowden} reported an enhancement in the \emph{XMM-Newton} background,
870:  especially at 0.5--1.0\,keV,
871:  concurrent with an enhancement in the solar wind
872:  measured by the \emph{Advanced Composition Explorer} (\emph{ACE})
873:  and other monitoring spacecrafts.
874: We thus obtained data
875: % from \emph{ACE}\footnote{\url{http://www.srl.caltech.edu/ace/ASC/}}
876:  from \emph{ACE}\footnote{http://www.srl.caltech.edu/ace/ASC/}
877:  and inspected the light curves 
878:  at the time of our observations (Fig.~\ref{fig:ace}).
879: %\placefigure{fig:ace}
880: We found that our observation of 0917+75 coincided with an episode
881:  of significant enhancement of the O$^{7+}$/O$^{6+}$ ratio in the solar wind. 
882: This factor conveniently explains the apparent line emissions in our spectrum 
883:  of 0917+75.
884: %\begin{figure}
885: %\label{fig:ace}
886: %\plotone{ace.eps}
887: %\caption{Proton flux and oxygen isotope ratios in the solar wind,
888: % measured by \emph{ACE} during the time of our observations.}
889: %\end{figure}
890: 
891: In contrast, for 1401$-$33, the count rates above about 10\,keV are
892:  well within the accepted quiescent levels.
893: As for energies below 1.0\,keV, the spectra are dominated by thermal
894:  emission from the halo of NCG~5419, swamping any spectral line
895:  due to the solar wind.
896: Besides, the \emph{ACE} data at the time of this observation show
897:  an average level of solar wind activity.
898: Thus, residual flaring should not have been a problem for 1401$-$33.
899: 
900: To obtain an upper limit of source flux,
901:  we recalculate the background level with multiple background regions
902:  at the vicinity of the original, and obtain the standard deviation $\sigma$
903:  of this sample.
904: We then set the upper limit as the $+3\sigma$ photon count level.
905: %, $\sqrt{\sum\mbox{photon count}} / \sum\mbox{exp. map}$.
906: %Using Equation~9 in \citet{1998PASJ...50..389H},
907: % which followed the treatment
908: % of synchrontron and inverse-Compton radiation
909: % in Rybicki and Lightman\cite{RandLbook},
910: Using Equation~11 in \citet{1979MNRAS.188...25H},
911:  Equation~9 in \citet{1998PASJ...50..389H},
912:  and radio measurements reported in \citet{HarrisSternWillis1993}
913:  and \citet{GossMcadamWellington1987},
914:  we obtained the lower limit of magnetic field strength
915:  as listed in Table~\ref{tab:resultsInferred}.
916: \fi
917: %\placetable{tab:resultsInferred}
918: %\input{tab04measuredParam}
919: %\input{tab06inferredParam}
920: